Mantle peridotite xenoliths from Archean cratons generally have high molar Mg/(Mg + Fe), or Mg#. The best known suites, from the Kapvaal and Siberian cratons, have high modal orthopyroxene (Opx). These high Opx compositions are probably not residues of partial melting. Less well known cratonic xenolith suites from Greenland and North America include high Mg# peridotites with much lower modal Opx. Such low Opx compositions could be residual from high degrees of polybaric, decompression melting, ending in the spinel lherzolite stability field at pressures of 30 to 20 kbar. This paper presents additional evidence that the great majority of both spinel- and garnet-bearing xenoliths are also residues of polybaric melting that ended at pressures ≤30 kbar. Where xenoliths record equilibration pressures > 30 kbar, this must result from tectonic transport of peridotites to greater depth after melting. Proposed mechanisms for producing the high Mg#, high Opx compositions include metamorphic differentiation of high pressure residues, mixtures of residual peridotites and high pressure igneous cumulates from ultramafic magmas, and addition of SiO₂ to low Opx peridotites via melt/rock reaction. This paper focuses on a positive correlation between Ni contents of olivine and modal proportions of Opx in mantle xenoliths, and uses this correlation to constrain the processes that produced high Mg#, high Opx cratonic mantle compositions. The observed correlation is probably not produced by partial melting, metamorphic differentiation, or formation of igneous cumulates. It can be produced by reaction between SiO₂-rich liquids (e.g., small degree melts of subducted eclogite) and previously depleted, low Opx peridotites. We propose a two step process. First, high Mg#, low Opx peridotites were created by large degrees of polybaric melting ending at pressures < 30 kbar. Later, these depleted residues were enriched in Opx by interaction with SiO₂-rich melts generated mainly by partial melting of eclogitic basalt and sediment in a subduction zone. Magmas modified by such a process could have formed a major component of the continental crust. Thus, this hypothesis provides a genetic link between cratonic upper mantle and continental crust.